New Leader on TSPCalc

[SurelyJame]

Just a heads up in case anyone wants to change their S allocation to the I fund today if you are following the leader (the mean, that is).
(Switch from 12503 to 16671)

[Hannah]

Std Dev: 27.81

[bamafamily]

Std Dev: 27.81

[hytid]

Recent trends have me thinking the risk * potential degree of a pullback from the S might be worth a temporary deviation from 7980.

That and the fact that one of the monthly strategies I consider recently made the jump to I have me moving to I instead of S this time.

*crosses all fingers as I hesitantly choose a new target to stay on temporarily*

[onerepmax]

here's what I consider the leader: 16556
mean SD mean/SD
2004-2016 27.46 2.25 12.2
2004-2017 26.94 2.87 9.39

I believe the 04-17 numbers will be much closer to the 04-16 numbers when the end of 17 rolls around.

[Hannah]

here's what I consider the leader: 16556
mean SD mean/SD
2004-2016 27.46 2.25 12.2
2004-2017 26.94 2.87 9.39

I believe the 04-17 numbers will be much closer to the 04-16 numbers when the end of 17 rolls around.

It looks as though 16556 is the same as 16517 from this point until the end of the year.

[bad70nova]

Wouldn't we want as low as possible STD Dev?

Am I missing something

[Hannah]

Wouldn't we want as low as possible STD Dev?

Am I missing something

Yes.

16517 has a slightly higher STD Dev than 16556. But 16517 YTD is 23.20% while 16556 is 20.20%. But since they both have the sames moves for the rest of this year, it really is a moot point now.

[Scooter]

Interesting. Thanks.

[onerepmax]

Wouldn't we want as low as possible STD Dev?

Am I missing something

yeah, my formatting didn't really come thru...

#16556 from 04-16
mean = 27.46
SD = 2.25
mean/SD = 12.2

#16556 from 04-17
mean = 26.94
SD = 2.87
mean/SD = 9.39

[Aitrus]

Wouldn't we want as low as possible STD Dev?

Am I missing something

I was wondering the same thing myself. I understood that STD Dev means the 1x % range of deviation from the mean, and sigma is multiples of that number. 1 sigma is 1x the standard deviation, while 6 sigma is 6x the standard deviation. There's no "finding" 6 sigma, it's already there. It's 6x what? A higher 1 sigma is bad because that means a higher 6 sigma.

To get 3 sigma (3 standard deviations from the mean), you have to multiply the STD Dev by 3. 3 sigma means that 93.3% of all returns will fall within that range. 6 sigma, which some have looked hard for, means that 99.999997% of all returns, present and expected future, will fall in that range. That means that 3.4 years out of a million will fall outside the range of expectation. If any of us is here in a million years to bemoan "those 3 years that got away", I would be very impressed.

Here's and example of how I understand STD Dev, using onerepmax's stats for 16556:

If 16556's 2004 - 2016 mean is 27.46% with a STD Dev of 12.2%. Since 16556's 3 sigma is 36.6%, then 93.3% of all returns, past and expected in the future, will fall within -9.14% to 64.06%.

A high CAGR with a low STD Dev would be preferable, in my mind, but I've seen a lot of posters who are crowing over finding STD Devs that are higher. To me, that means that the potential highs are better, but so are the potential lows.

Unless I'm misunderstanding the relationship between STD Dev and sigma. Somebody feel free to correct me if I'm wrong.

[Ofsthun01]

I get that having a higher mean/SD (sigma) is a good thing. But at what point does a higher sigma become irrelevant?

Sigma (σ) is the number of standard deviations away from the mean. It is determined by dividing the mean by the standard deviation. From a risk perspective, you can determine how risky a strategy is by the sigma calculation:
1σ = 32% of the returns will be outside 1σ (higher or lower) 16% chance of a loss
2σ = 4.5% of the returns will be outside 2σ (higher or lower) 2.25% chance of a loss
3σ = 1 year out of 370 years may be a loss
4σ = 1 year out of 15,787 years may be a loss
5σ = 1 year out of 1,744,278 years may be a loss
6σ = 1 year out of 506,797,346 years may be a loss
7σ = 1 year out of 390,682,215,445 years may be a loss

Personally, anything that's a 3/4 seems pretty safe. Once you start getting into 8s, 9, 10s, etc you're sacrificing CAGR and a better mean for the 1 in a quintillion chance it might a negative return. Is that really worth it?

[WxMan0523]

There's a problem with ALL of this statistical mumbo jumbo. NONE of these numbers are statistically significant! Let's say that past returns ARE a predictor of future success. Even so, we have 13+ years of data to work with here. You need ~30 years worth of data to put any sort of weight behind the std dev/sigma values being spit out by the any of these daily seasonal offerings. Be careful. TomComm's sigma numbers are correct ONLY WHEN the values are of statistical significant. We're not even halfway there yet.

[sjonesuspp]

Actually the leader is Cgordon

[fordest]

Aitrus, you seem to be confusing his posted numbers. The 12.2 is the Mean/STD. The standard deviation is only 2.25. Run your numbers agains using 2.25.

A higher Mean/standard deviation is desirable as you mentioned. That's that the 12.2 is. And that's the sigma.